castle: role of selection in evolution 381 



In the case of certain characters in guinea-pigs I have re- 

 peatedly attempted modification of a racial character by selec- 

 tion within an inbred race, without success. Thus a very dark 

 form of Himalayan albino, after a certain amount of improve- 

 ment by selection, could not be further darkened to any appreci- 

 able extent. A race selected simultaneously for large size and 

 for small size showed so little change that the experiment was 

 abandoned after a few generations. No indication was forth- 

 coming that we could thus ever approach in size either the small 

 wild Cavia Cutleri of Peru, or the large races of guinea-pig kept 

 in captivity by the natives of the same region. Yet evolution 

 had in some way evidently produced these divergent conditions 

 from a single original source. The changes were probably too 

 slow to be observable in the life time of one observer. 



On the other hand, certain characters of guinea-pigs, rabbits, 

 and rats have been found to respond readily to selection in a 

 particular direction. This is notably true of color patterns 

 which involve white spotting. A selection experiment with 

 hooded rats selected simultaneously in plus and minus directions 

 has produced one race which is black all over except a white 

 patch of variable size underneath, and another race which is 

 white all over except for the top of the head and the back of the 

 neck, which are black. The races do not overlap at all and have 

 not done so for many generations, though they still continue to 

 diverge from each other as a result of continued selection. 



In similar experiments with Dutch marked rabbits it has been 

 found possible by selection to increase or decrease the amount of 

 white at will. In a series of such rabbits ranging from nearly 

 all black to nearly all white, stages far enough apart to be cer- 

 tainly identifiable behave as Mendelian allelomorphs in crosses, 

 but regularly emerge from such crosses in a slightly modified 

 form, the whiter stages having been darkened and conversely 

 the darker stages whitened. The principle of the pure line 

 manifestly does not apply to these cases. White spotting is 

 apparently a character which from its nature fluctuates con- 

 stantly, such fluctuations having, to some extent at least, a 

 genetic basis, since continuous selection invariably produces a 



